There has been a lot of talk lately about the potential of the Bradley Effect having potential to turn the election. While poking around some numbers at Pollster.com, I found something that seemed unusual.
When live telephone polling is used, and a person has to tell another live person for whom they are voting, the race is indeed tightening. The interesting thing is when you filter the results, and take out those 'live' polls, and go only for automated telephone and internet polling, you get vastly different results. So, much like in the voting booth, when people are in private, and not giving their choice to a live person, but rather to a machine, the complete opposite of a Bradley effect seems to be in play.
See charts below the jump . . .
***Note, if the charts are not rendering properly for you, follow the link below each chart.***
First up, the overall numbers from September 1st to the present date, showing all polls. McCain is showing slight gains, while Obama is showing a small, nearly statistically insignificant drop in the last two weeks.
<script></script>
Chart 1
Second up, I factor out the 'live' polling and allow in only automated phone and internet polling, by this methodology, both candidates have been gaining at a slow and steady pace, something logical, since undecideds are splitting up into both McCain and Obama voters.
<script></script>
Chart 2
Third, I factor out the 'less reliable' internet polling, and go strictly with automated phone polling, I realize now we are not looking at a lot of polls, but the numbers here are vary favorable for Obama, as the trend line shows he is in fact widening his lead while McCain is losing ground.
<script></script>
Chart 3
So, there is my interpretation. I think there may be some people out there that are afraid to tell people they are actually voting with their best interests, because they are afraid their peers will not understand. However, when they are in private, as they will be on November 5th, they are voting differently.